Foundational cases: free-recall review
Free-recall prompts across the foundational case studies. Reconstruct each design from memory — the KV store's three blocks, the shortener's read path, the crawler's frontier and dedup, the queue's log — before revealing the model answer.
Retrieval beats re-reading. For each prompt, reconstruct the full design from memory — the components, the numbers, the trade — before you open the model answer. The effort of rebuilding the architecture is what makes the patterns transfer to the next case you’re handed.
Rebuild each design without looking back: the three ideas a Dynamo-style store stacks, why a URL shortener is a read problem, what the crawler frontier and dedup layers each do, and how a partitioned log delivers ordering, durability, and replay.
- 01Reconstruct the Dynamo-style KV store: the central trade and the three building blocks.
- 02Why is a URL shortener a read problem, and how do you mint the code?
- 03Reconstruct the crawler: what the frontier and the two dedup layers each do.
- 04Reconstruct the message queue: log vs mailbox, partitions, ISR, and delivery semantics.
If you could rebuild each design from memory, you hold the unit’s spine: the four cases are the same moves reshaped. The KV store picks availability over consistency and stacks a hash ring, a quorum (R + W > N), and versioning, with gossip and anti-entropy underneath. The URL shortener is a ~100:1 read problem solved with a counter/KGS for codes and a cache for the redirect, with 302-vs-301 trading analytics for load. The crawler is a two-stage frontier (priority + politeness) plus two dedup layers (bloom filter for URLs, content hash for bytes), made continuous by freshness-driven recrawl and hardened against traps. The message queue is an append-only partitioned log: partitions for parallelism and ordering, the ISR with acks=all for durability, and offset-commit timing for the delivery semantics. The transferable lesson is that every new “design X” prompt decomposes into requirements, an estimate that picks the shape, a few hard components, and the bottlenecks you name before you build.
Something unclear?
Ask a question about this lesson. Questions are anonymous and go straight to the author to make the lesson better.